Multi-stage Alzheimer's disease diagnosis using Diffusion Tensor Imaging and Cost-Sensitive Machine Learning
Early and accurate diagnosis of Alzheimer's disease (AD) stages, including Early and Late Mild Cognitive Impairment (EMCI, LMCI), is crucial for intervention. This study leverages Diffusion Tensor Imaging (DTI) metrics from 57 brain regions to classify AD progression and differentiate from heal...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
ICI Publishing House
2025-06-01
|
| Series: | Revista Română de Informatică și Automatică |
| Subjects: | |
| Online Access: | https://rria.ici.ro/documents/1377/art._Khazaee_Mohammadi.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Early and accurate diagnosis of Alzheimer's disease (AD) stages, including Early and Late Mild Cognitive Impairment (EMCI, LMCI), is crucial for intervention. This study leverages Diffusion Tensor Imaging (DTI) metrics from 57 brain regions to classify AD progression and differentiate from healthy controls (HC). The proposed method, which incorporates various machine learning models, Bayesian hyperparameter optimization, 10-fold cross-validation, and cost-sensitive learning, achieved a high test accuracy of 90.4%. Feature ranking consistently identified Axial Diffusivity in the left uncinate fasciculus as a key biomarker, alongside important contributions from the sagittal stratum and hippocampal cingulum. Our findings demonstrate the significant potential of the DTI-derived features combined with optimized machine learning for enhancing multi-stage AD diagnosis and understanding the underlying neuropathological mechanisms. |
|---|---|
| ISSN: | 1220-1758 1841-4303 |